French loanwords in Vietnamese: the role of input language phonotactics and contrast in loanword adaptation

Studies on loanword phonology show that the restrictions of native (L1) phonology play a crucial role in reshaping foreign words in adaptation. Studies have also found, however, that there are many aspects of loanword adaptation that are not explained by the requirements of native phonology alone (See Kang (2011) for a recent overview). A considerable body of work has accumulated to show that such unexpected adaptation patterns may be explained when the phonetic details of the source language (L2) and the native language (L1) sound structure are taken into account. Specifically, it has been proposed that adapters select the native language structure that is perceptually most similar to the borrowing language input as the optimal mapping. This selection may be achieved in the process of speech perception proper (Boersma & Hamann 2001; Peperkamp et al. 2008) or in the process of a mapping which is mediated by speakers’ knowledge of relative phonetic similarity between strings of sounds (Kenstowicz 2007; Steriade 2008; Yun To appear).

Relatively less attention has been paid to the role of knowledge of the source language phonology that adapters bring to adaptation. LaCharité & Paradis (2005) proposed that the adapters are competent bilinguals with native-like knowledge of the input language phonology and the phonological structure of the source language serves as input to the adaptation process, rather than the surface phonetic forms of the source language. But, others assume a more nuanced view of the role of input language phonology and propose that the adapters’ knowledge of source language phonology can be one of various factors that affect the outcome of adaptation and the adapters’ knowledge may not be native-like (de Jong & Cho 2012; Ito 2014; Kang 2010; Smith 2009). Kang (2010) and de Jong and Cho (2012) in particular proposed that while phonetic similarity plays a primary role in shaping adaptation, the variability inherent in cross-language perception is modulated by speakers’ knowledge of input language categories, exerting a regularizing pressure on the adaptation over time. In this paper, we provide two case studies of French-Vietnamese vowel adaptation where the Vietnamese adapters’ knowledge of L2 (French) phonology seems to play a crucial role. The study also makes an empirical contribution by providing a detailed description of French-Vietnamese vowel adaptation.

2 Data

Data for the study consist of over 1,000 French loanwords in Vietnamese. The list is primarily drawn from three dictionaries (Lê et al. 1988; Nguyễn et al. 1998; Nguyễn 1986[1975]) and is supplemented with commonly used forms that the second author is aware of. Additional data were added from Huynh (2010)’s loanword list which is itself collected from various published sources and dictionaries and checked for usage and accuracy with Vietnamese informants from Saigon and Hanoi. The second author also elicited productions of around 30 commonly used loanwords using a picture naming method from 14 native speakers residing in Vietnam in the summer of 2013. The fieldwork data are not included in our analysis but they serve to confirm the general validity of the French-Vietnamese sound correspondences we observe in our written data sources. Lexique 3.80 (New et al. 2001)¹ is used as a source for French phonological transcription and the phonological transcription of the Vietnamese forms is aided by vPhon, a Python script that converts Vietnamese scripts to IPA transcriptions (Kirby 2008). Previous phonological descriptions of French loanwords in Vietnamese include Barker (1969), Huynh (2010) and Phillips (1975).

In this section, we provide background information on the phonology of Vietnamese (Emerich 2012; Hoa 1965; Hwa-Froelich et al. 2002; Kirby 2011; Phạm 2006; Phạm To appear; Tang & Barlow 2006; Thompson 1965) and French (Féry 2003; Fougeron & Smith 1999; Storme 2015; Strange et al. 2007; Tranel 1987; Walker 2001). The information on Vietnamese is based on the Northern (Hanoi) dialect.² The inventories of vowels and consonants of the two languages are presented in (1)-(4).

Vietnamese has nine monophthongal vowels and three diphthongal vowels. Two non-high central vowels, /a/ and /ɤ/, show length contrast and these contrasts are also accompanied by spectral differences, with the differences attested more consistently for the /ɤ/-/ɤ̆/ pair than the /a/-/ă/ pair (Emerich 2012; Kirby 2011). The short vowels /ă/ and /ɤ̆/ only occur in closed syllables. Also there is a marginal length contrast of /ɛ/ vs. /ɛː/ and /ɔ/ vs. /ɔː/; the long vowels /ɛː/ and /ɔː/ only occur in syllables closed with a velar coda. Vietnamese coda consonants are limited to voiceless stops [p, t, c, k, k͡p], nasals [m, n, ɲ, ŋ, ŋ͡m] and approximants [j, w]. Labial codas [m, p] can occur with all vowels, except for /ɯ/. In the Northern dialects, dental codas [t, n] can occur with all vowels although they are rare after /ɯ/. Palatal, velar, and labial-velar codas are in quasi complementary distribution, as summarized in (5). The overall pattern is that dorsal codas assimilate to the place of articulation of the preceding vowel, except for /ɛː/ and /ɔː/. Palatals ([c] and [ɲ], or fronted velars [k̟] and [ŋ̟]) are found following a front vowel, labial-velars ([k͡p] and [ŋ͡m]) are found following back rounded vowels and velars ([k] and [ŋ]) are found following central vowels. Velars are also attested following long mid vowels /ɛː/ and /ɔː/, as in xẻng [sɛːŋ] ‘shovel’ and xoong [sɔːŋ] ‘saucepan’. The occurrence of palatals following a low central short vowel /ă/ seems out of place and many analyze the vowel as underlying /ɛ/ which is centralized in a palatal context (See Pham to appear and references therein).⁴

French has twelve monophthongal oral vowels and four nasal vowels. The contrasts between /a/ and /ɑ/ and between /ɛ̃/ and /œ̃/ are marginal. Unlike Vietnamese, French has contrasts between front unrounded and front rounded vowels and also a series of nasal vowels. Similar to Vietnamese, French contrasts tense (=higher) and lax (=lower) mid vowels. These mid vowels of French are subject to a phonotactic restriction known as the Loi de Position (LP) (Féry 2003; Storme 2015); generally, the lax vowels /ɛ/, /œ/ and /ɔ/ are found in closed syllables and the tense vowels /e/, /ø/ and /o/ are found in open syllables. This restriction holds fairly systematically in non-final unstressed syllables.⁵ But, in final stressed syllables, there are exceptions; /e/ and /ɛ/ contrast in word-final (=stressed) open syllables while /o/ and /ɔ/, and /ø/ and /œ/ contrast in word-final closed syllables. But, even in final closed syllables, only /o/ is found before /z/ and only /ɔ/ is found before /ʁ/, /ɲ/ or /ɡ/. The pattern in the standard French is summarized in (6). It is noted that /ɔ/ is realized with a more centralized quality phonetically before /ʁ/ (Fougeron and Smith 1999). The Loi de position is also subject to dialectal variation (Féry 2003; Walker 2001).